Great, another reason for the increasingly bloated plaintext chat logs, just what I wanted. Because nothing says "conversational AI" like extra overhead from compressing agent context. https://github.com/Compresr-ai/Context-Gateway